Author includes comments (irrelevant to code to some extent) and braces (enforced by compiler) and particulars of code style into LOC and calls it unreliable.
I think it is not quite honest.
The rules themselves vary between languages and, frankly, contemporary C++ is a different language than C++ circa 1997. Often, use of C++ is different between parts of projects and it is the case for contemporary Java too.
Should we develop different ABC rules for different flavors of C++ and/or Java?
For the lazy like me who are interested in the topic but googling for it is too exhausting:
https://en.wikipedia.org/wiki/ABC_Software_Metric
Long story short: instead of counting lines of code, the ABC software metric accounts for
A) assignments ,
B) branches,
C) conditionals.
I would argue that the ABC metric nowadays fails to do a good job as a software metric, mainly due to the inception and prevalence of declarative frameworks and tools.